Research On Calibration And Positioning Of A Spherical Artificial Compound Eye

As an important method in three-dimensional measurement field,visual positioning has been widely used in industrial inspection,medical treatment,visual effects and security monitoring and so on.With the expansion of application field,the requirements of the vision system are getting higher and higher,such as large field-of-view,miniaturization and high precision.The compound eye of many arthropods in nature has attracted a lot of attention of researchers in visual measurement field by their advantages of compact structure,large field-of-view and high motion sensitivity.Therefore,our research group design and fabricate a new type of artificial compound eye positioning device.In order to achieve the calibration and positioning of the designed compound eye,based on the characteristics of the system,this thesis analyzes the distortion characteristic of the system imaging,and proposes an calibration scheme based on virtual double spherical surface.The key problems are explored in calibration and positioning process.The three-dimensional positioning task of the object of space object is completed and the positioning accuracy is evaluated.The main contents are as follows:(1)The artificial compound eye device is described in detail from the aspects of mechanical structure,optical design and image acquisition system.The distortion of some eyelets imaging channels is analyzed by experiments,and the importance of the calibration to the compound eye system is clarified.On the basis of investigating various nonlinear camera calibration methods,the calibration method is discussed based on the compound eye characteristics of large field-of-view,many eyelets and complex distortion.We propose a calibration scheme based on virtual double sphere,which realizes the calibration by constructing the mapping relationship between image points and corresponding incident rays.And we have established a calibration mathematical model,set up a calibration platform consisting of single LED,two-dimensional rotary table,one-dimensional translation table and automatic control software.(2)Based on the calibration principle,the calibration step is reasonably planned and the calibration platform is adjusted at first.Then,based on the feature of spot image,the extracting step and algorithm of spot center are determined.In order to ensure the uniformity distribution of the space target,according to the rotary table rotation feature,the regular icosahedron subdivision scheme is introduced,and the target location is reasonably planned.For multi-channel simultaneous imaging in the same image,the matching method between the spot and the eyelets is explored by integrating the rotating angle and the eyelets position in the same polar coordinate system.In order to establish the nonlinear mapping relation,the method which is most suitable for the system is selected in the comparison and analysis of three mapping methods.Finally,the calibration experiment is carried out for the compound eye,and the nonlinear mapping relationship between the image point and the incident ray is established.(3)Based on the principle of binocular vision and system calibration,we probe into the theory of compound eye positioning,and study the matching method of the channel which the spot belongs to in the positioning process.Using compound eye to measure the three-dimensional point,the results show that the target relative positioning error of the compound eye device is better than 0.5%in 60° field,and the mean square root error of the positioning angle is about 1.96 mrad.Finally,We have measured and reconstructed the simple surface,which demonstrates the practical application value of the designed compound eye.